Plant recognition based on identification of leaf image using image processing / Nor Silawati Sha’ari

It is challenging task to analyze plant leaf images by a layman because there are very minute variations in some a plant leaf image and the large data set for analysis. It is also a quite difficult to develop an automated recognition system which could process a large information and provide a correct estimation. In this paper, by using the database available in the internet and using Neural Network (NN) as training algorithm, plant recognition based on leaves image would be developed. Image processing techniques are used to extract the leaf feature from histogram of the leaf image. These extracted features are used as inputs to neural network for classifying the plants. NN such as Artificial Neural Network (ANN) and K-Nearest Neighbor (KNN) is trained in developing a classification system for agriculture purpose. ANN and KNN is applied to solve the problems in image analysis, pattern recognition and classification. For the ANN, Multilayer feed-forward networks are trained using Back Propagation (BP) learning algorithm and for the KNN, is used the most common distance which is Euclidean. The main objective is to developing a classification system for agriculture plant by image pre-processing, feature extraction, network training and classification. Under the current research, 80 leaves from 4 kinds of plants were collected. Out of these, 40 leaves were trained. The 40 leaves testing samples were recognized with 97.5% accuracy on ANN classify and 87.5% accuracy from KNN classify. The analysis and simulation of the classification of leaves image from different characteristics are tested are implemented in MATLAB.